The role
- COMP
- $145K - $275K
- EQUITY
- RSUs
- LOCATION
- Seattle
- WORKPLACE
- Hybrid
- EXPERIENCE
- 4 - 10 years
- VISA
- None
- STACK
- Python, C/C++, Rust, Java, Langchain
- INDUSTRY
- Software Development, Government, Defense, Data
/roles — ROLE_191
Software Engineer - Edge AI Systems
Large public software company deploying AI and data platforms for defense and government missions
The company
Large, publicly traded software company whose data integration and analytics platforms help governments, militaries, and major enterprises run critical operations.
- STAGE
- public-scale
- FUNDING
- public-scale
- TEAM
- 1,000+ people
- BACKING
- backed by Goldman Sachs
JD — the work
About the role
You would join a newly formed team putting advanced AI onto field-deployed edge hardware for defense customers. The team trains, evaluates, and deploys LLM agents that run on constrained devices and hook into drones, sensors, and effectors, so capable AI can operate in the field without depending on a connection back to the cloud. You would work alongside AI, LLM, and systems specialists on problems where reliability under pressure really matters.
What you'll do
- Train and fine-tune compact LLMs suited to tactical defense use cases
- Tune model and runtime performance across CPUs, GPUs, and dedicated accelerators
- Architect multi-agent setups for a range of mission scenarios
- Connect agents to drones, sensors, and effectors in the field
- Benchmark agents for capability, speed, and reliability
- Build fail-safe, dependable systems for high-stakes environments
What they're looking for
- 5+ years building software professionally, including 2+ years owning system design and architecture
- Hands-on experience building and shipping LLMs and agents with PyTorch, Hugging Face Transformers, LangChain, or vLLM
- Deep fluency in Python, Rust, or C++
- A bachelor's, master's, or PhD in CS, physics, mathematics, or similar
- Current or obtainable US security clearance (you must be eligible and willing)
Nice to have
- Defense experience, particularly with weapons or unmanned vehicle systems
- Familiarity with edge optimization tooling such as vLLM, TVM, TF Lite, Ollama, or Mojo
- Experience optimizing models and runtimes for edge devices
- Track record running robust production systems